#emc-devel | Logs for 2010-12-14

[02:10:21] <jepler> JT-Shop: I'll try that.
[02:10:57] <jepler> JT-Shop: do you think errors behave different than debug?
[04:55:40] <dimas_> dimas_ is now known as 77CAAH103
[05:43:59] <dgarr> jepler: i've seen the errors in trying to remove axis notifications. i think they occur when there has been a mix of info and error notifications. maybe this is the problem: http://www.panix.com/~dgarrett/stuff/0001-axis-notification-button-command-corrected.patch
[14:38:27] <psha> they deny you running logger on emc-dev channel? :)
[14:38:34] <Jymmm> No, it's theirs unless you wrote it
[14:38:36] <SWPadnos> see how I craftily said that while the loggers were off :)
[14:39:09] <SWPadnos> one of the terms says thou shalt not run bots and things
[14:39:17] <Jymmm> ah
[14:39:34] <psha> heh, if it's not working when you sleep it seem that you are logging manualy :)
[14:39:43] <psha> so no violations :)
[14:39:52] <SWPadnos> psha, I do not understand the command
[14:39:57] <SWPadnos> :)
[14:40:43] <psha> what command?
[14:40:56] <psha> returning to my panels :) what widgets will be useful/nice in touchy?
[14:40:58] <SWPadnos> that was my attempt to look like a bot that doesn't know what to do
[14:41:22] <psha> then i've simulated bot who don't understand whats happening around :)
[14:43:04] <skunkworks> so - I just found out I have an issue with my pallet change code. (right now it is a coulple of buttons on the pyvcp) I issue mdi commands to move the table. This won't work in gcode if I want to do it automatially. (auto)
[14:44:00] <SWPadnos> subroutine
[14:44:06] <SWPadnos> O<palletchange> call
[14:44:19] <skunkworks> oohhhh - that is doable.
[14:44:39] <skunkworks> have to look at that.
[14:44:42] <SWPadnos> ah, here it is: http://wiki.dreamhost.com/Cron_Jobs_%26_Persistent_Processes#What_is_your_persistent_.28background.29_process_policy.3F
[14:44:54] <SWPadnos> the first item in the list :)
[14:44:56] <skunkworks> (I don't know if we need fully auto pallet changes any time soon)
[14:45:16] <skunkworks> dad is retired - he can push the button.
[14:45:21] <SWPadnos> heh
[14:46:18] <Jymmm> SWPadnos: But... # Running programs via cron is allowed provided you don't use excessive system resources.
[14:46:36] <Jymmm> SWPadnos: so cron an irc CLIENT
[14:46:50] <SWPadnos> the loggers are PERL-based IRC clients
[14:47:12] <SWPadnos> and note what they say in the second sentence of the "May I run a cron job?" section
[14:47:14] <Jymmm> so there ya go, just corn em
[14:47:47] <SWPadnos> it is a background process, and is specifically listed (first!) as the exact type of prohibited process they're talking about
[14:48:13] <SWPadnos> so let's stop discussing it, lest they notice :)
[14:49:14] <psha> maybe add in cron some form of restarting?
[14:49:24] <SWPadnos> we tried that
[14:49:48] <SWPadnos> also tried a change to the script to re-run it if it dies - we ended up with two loggers every time
[14:49:56] <SWPadnos> or more :)
[14:51:45] <psha> it was cool time!
[14:56:42] <jepler> dgarr: assign default values to vmajor/vminor in case the version is not in the expected form?
[14:56:49] <jepler> double vmajor=0.0, vminor=0.0;
[15:01:50] <cradek> psha: I'd be happy to see gremlin in a touchy example, but I think jog buttons are bad
[15:02:22] <cradek> it is on purpose that touchy does not have jog buttons
[15:03:04] <Jymmm> O_o
[15:03:43] <jepler> Jymmm: cradek means onscreen jog buttons
[15:03:49] <Jymmm> oh
[15:04:01] <psha> http://psha.org.ru/tmp/touchy-graph.png
[15:04:12] <psha> while doing snowflake
[15:04:14] <jepler> touchy is only part of a panel that will also have physical jog switches/jog wheel
[15:04:36] <Jymmm> jepler: a pendant of sorts?
[15:04:49] <psha> cradek: ok :) i won't add jog panel into touchy.ini
[15:05:25] <jepler> psha: do we have your localization of touchy in our git?
[15:05:53] <psha> sure, it's yours localisation :)
[15:06:03] <jepler> it must not be mine personally
[15:06:13] <psha> i mean it's not mine :)
[15:06:31] <psha> i think it's derived from axis but not sure
[15:07:24] <jepler> at least gtk can render cyrlic letters pretty well. tk8.4 at least looked TERRIBLE: http://media.unpythonic.net/axis-files/translations/axis-ru.png
[15:08:09] <psha> jepler: axis is rendering letters good if it founds proper font
[15:08:17] <jepler> yeah, tk8.5 on ubuntu 10.04 looks considerably better
[15:11:30] <psha> cradek: btw may you clarify what's bad with jogging in touchy? for abort/run in understand that it's horrible to have them on touchscreen
[15:12:00] <psha> but jogging is not a crime or i'm missing something?
[15:12:02] <cradek> psha: if my finger is wet, my touchscreen does not see when I stop touching
[15:12:39] <psha> so you may crush your tool into material?
[15:13:06] <SWPadnos> even if there's no damage, the machine should stop when you let go of the button
[15:13:25] <cradek> yes in this case, it would keep moving after I let go
[15:13:31] <SWPadnos> there are several kinds of touchscreen, and different things will screw up the different types
[15:13:33] <cradek> that is really bad, maybe the worst possible thing :-)
[15:13:47] <psha> i see now...
[15:13:52] <cradek> the thing all touchscreens have in common is they are unreliable!
[15:14:10] <SWPadnos> the same could happen with spindle speed +/-, feed override, etc
[15:14:19] <psha> but for mouse-and-keyboard usage it's ok?
[15:14:22] <SWPadnos> (if there are continuous up/down buttons)
[15:14:27] <cradek> spindle speed +/- is a certain amount per click
[15:14:32] <SWPadnos> mouse buttons usually release when you let go of them
[15:14:36] <cradek> feed override is controlled by the wheel
[15:14:55] <cradek> yes mouse and keyboard are generally reliable - touchscreens are not
[15:15:01] <SWPadnos> heh
[15:15:03] <jepler> dgarr: I'll add in the initialization and push
[15:15:18] <SWPadnos> clicky buttons are so much easier to sense failures on
[15:16:37] <SWPadnos> (with your fingers that is, not with software)
[15:17:52] <psha> i see... but i think that for some people touchy may serve as axis replacement...
[15:18:36] <SWPadnos> yep. people with a touchscreen and a physical jog wheel :)
[15:19:13] <psha> :)
[15:19:30] <SWPadnos> if you like the format of touchy, but want to be able to use it with a mouse and keyboard, you could make a fork called "clicky" :)
[15:19:49] <psha> :)))
[15:19:50] <cradek> psha: I agree some people may try to use it with a mouse - but it is very awkward if so
[15:20:15] <cradek> ("typing" mdi commands with a mouse when you already have a keyboard is stupid)
[15:20:26] <psha> yes, that's not fine ;)
[15:20:35] <psha> but selecting .ngc file is nice...
[15:20:40] <psha> if you set correct paths
[15:21:12] <cradek> yes I like how easy that is
[15:21:37] <Jymmm> Could a mouse scroll wheel be used for jogging?
[15:22:02] <cradek> you could use a mouse wheel in place of the jog wheel I guess, but it would be awful
[15:22:03] <SWPadnos> most likely, if you want to modify some code
[15:22:07] <psha> hm, is touchy recursive in it's search for ngc files?
[15:22:12] <cradek> no
[15:22:41] <psha> is it 'feature'? :) or just not bothered with it?
[15:23:08] <cradek> I did not want to make a directory browser because I think it is unnecessary and overly complicating
[15:23:19] <cradek> I think recursive search is also unneeded
[15:23:48] <cradek> there is a directory. ngc files you place there are "in the control". nothing else is needed IMO
[15:24:18] <psha> good point
[15:24:29] <cradek> you can even move files around with the touchscreen using the gnome stuff, if you really want to
[15:24:29] <Jymmm> what about ngc filed into folders per project/revision/component?
[15:26:01] <cradek> Jymmm: unnecessary complication IMO
[15:26:37] <cradek> Jymmm: you just load the project's ngc files "into the control" when you do other setup things like measuring tools, mounting fixtures, etc
[15:30:58] <cradek> touchy is really meant to be built into a machine's front panel - if you have a computer on a desk with keyboard/mouse/monitor, AXIS is much better
[15:31:29] <cradek> I think people want to merge all the features of each into both - I think that is a mistake - they are very different because they serve different purposes
[15:32:26] <psha> touchy may serve as core for gtk axis i think...
[15:32:31] <psha> at some point
[15:32:43] <cradek> yes possibly
[15:33:25] <SWPadnos> hmmm. on 10.04, after apt-get build-dep emc2 and install emc2-dev, lyx can't build several (maybe all) documents, because "book.layout" is missing. any ideas?
[15:33:43] <cradek> ew
[15:33:44] <psha> btw there is way to embed gladevcp panels without xembed
[15:34:28] <psha> i've used it in camview
[15:35:28] <SWPadnos> and I will say the Phenom II 1090T is friggin fast. 14 seconds to build EMC sim (no drivers of course)
[15:35:39] <SWPadnos> about 27 seconds with non-functional docs :)
[15:36:13] <SWPadnos> I'm sure the SSD helps there too
[15:37:11] <psha> heh, it's fast...
[15:39:04] <psha> somewhat in 2004 company where i was working bought new server for build process... it was based on xeon and costs >>1k$... and after it was configured it appears that my comp on athlon 64 bought for ~200$ was building our project significantly faster :)
[15:39:18] <SWPadnos> heh
[15:39:31] <SWPadnos> this one was $3k, so don't feel too bad :)
[15:40:28] <psha> i try to fit in ~350-400$ for new comps and ~200 for upgrades ;)
[15:41:32] <jepler> SWPadnos: how much user+sys CPU is that?
[15:42:29] <SWPadnos> user: 1:59.33, sys 0:14.83
[15:42:32] <SWPadnos> or close
[15:42:55] <SWPadnos> that was for the failed build with docs, but it included depends and build (I did make clean first)
[15:45:40] <jepler> this is weird
[15:45:40] <jepler> 246.93user 68.92system 1:03.65elapsed 496%CPU (0avgtext+0avgdata 267760maxresident)k
[15:45:47] <jepler> I have a 4-core system, so how do I get 496% CPU?
[15:45:58] <SWPadnos> ?
[15:46:00] <SWPadnos> dunno
[15:51:41] <psha> jepler: extra core as a gift to each 1000ths customer!
[15:53:48] <jepler> SWPadnos: so assuming we're building comparable software (emc2 git master) with comparable compilers (ubuntu 10.04) and comparable targets (amd64), that CPU of yours does twice as much work per core-second compared to my Phenom 9600
[15:58:21] <SWPadnos> I'll pull - this is probably a couple of months old
[15:59:01] <ries_> ries_ is now known as ries
[15:59:01] <jepler> I hope we haven't doubled the time to compile within the last few months
[15:59:04] <SWPadnos> heh
[17:38:23] <SWPadnos> jepler, after git pull, make takes ~21 seconds, 47 seconds with docs (which all works now ?)
[17:38:51] <SWPadnos> using make clean then time make -j12
[17:39:09] <SWPadnos> so it did increase by ~50%, or something else changed
[17:43:08] <psha> cradek, jepler: any time for merge?
[17:43:37] <cradek> I can do it
[17:43:42] <psha> col
[17:43:43] <psha> col
[17:43:44] <psha> cool
[17:43:45] <psha> :)
[17:43:54] <psha> branch gladevcp
[17:44:16] <cradek> what's the url?
[17:44:24] <psha> today merge consists of some widgets, persistance support for panels and glib bindings
[17:44:26] <cradek> oh wait, I have it
[17:44:30] <psha> git://psha.org.ru/psha/emc2.git
[17:44:46] <psha> also there is example config for touchy
[17:45:13] <psha> i've not included jog and run buttons as you don't like this idea :)
[17:46:06] <psha> to look at gremlin panel run emc configs/gladevcp/touchy.ini
[17:46:10] <cradek> are you sure you have everything on that branch?
[17:46:21] <cradek> I don't see any config changes
[17:46:52] <psha> d16e9d0
[17:47:09] <psha> oops
[17:47:17] <psha> now it's there
[17:47:49] <cradek> aha
[17:48:23] <psha> forgot to push final version
[17:49:57] <psha> there is one change in hal.py that allows GLib-style wrappers for pins
[17:50:05] <mhaberler_> the wiki will be a jour later in about 2hours
[17:50:07] <psha> everything else is gladevcp specific
[17:52:24] <psha> meter and graph widgets are just seeing light of the day so there may be some issues with them
[17:52:35] <cradek> emc configs/gladevcp/touchy.ini fails with several problems
[17:52:51] <cradek> there is no tool table, and I get ImportError: No module named gtkgl.widget
[17:53:39] <psha> gtkgl is needed for gremlin
[17:54:18] <psha> i've copied touchy.{ini,hal} and core_sim.hal from sim
[17:54:28] <psha> is there also tooltable needed?
[17:56:05] <cradek> hmm, now I have ImportError: No module named Xlib
[17:56:42] <cradek> did you add these new dependencies to the deb stuff?
[17:56:50] <psha> not yet
[17:57:05] <psha> xlib is dep for gladevcp to embed into parent window
[17:57:11] <cradek> aha, now it runs
[17:57:14] <psha> for gtk it's not needed though
[17:57:20] <psha> but for axis it's crucial
[17:57:47] <psha> gtkgl will be mentioned in docs/added to deps
[17:58:25] <cradek> Plot tab is really neat
[17:59:00] <psha> added simpocket.tbl
[18:00:31] <cradek> (minor thing) I wish left (touchscreen) button would pan in the four view gremlin/touchy
[18:01:01] <cradek> not sure how those should be controlled in the long run - seems like you need some stuff to make gremlin navigable by touchscreen
[18:01:22] <cradek> but this all looks great, thanks for all this work
[18:01:56] <psha> gremlin tab is only example
[18:02:00] <CIA-41> EMC: 03cradek 07master * r66e23f8e3e49 10/lib/python/gladevcp/led.py: gladevcp: Fix dark color broken after hardy backporting
[18:02:03] <CIA-41> EMC: 03cradek 07master * r71f6ea46b8a7 10/lib/python/gladevcp/led.py: gladevcp: Cleanup HAL Led a bit
[18:02:05] <CIA-41> EMC: 03cradek 07master * rc97b4087ae2f 10/ (20 files in 14 dirs): Merge remote branch 'origin/master' into gladevcp
[18:02:06] <CIA-41> EMC: 03cradek 07master * r4d0f640099ae 10/lib/python/gladevcp/hal_bar.py: gladevcp: Check Bar text_template property on set
[18:02:07] <CIA-41> EMC: 03cradek 07master * rd5114846cffc 10/lib/python/gladevcp/ (hal_bar.py hal_gremlin.py hal_widgets.py): gladevcp: Added license headers to new files
[18:02:08] <CIA-41> EMC: 03cradek 07master * rac04428a7adf 10/src/hal/user_comps/gladevcp.py: gladevcp: fix -d debug flag
[18:02:09] <CIA-41> EMC: 03cradek 07master * rbfae7b7925e3 10/lib/python/gladevcp/hal_widgets.py: gladevcp: Improve HAL_ComboBox widget
[18:02:10] <CIA-41> EMC: 03cradek 07master * r2e5e6f6e6672 10/lib/python/gladevcp/ (hal_meter.py hal_python.xml hal_pythonplugin.py): gladevcp: Add HAL Meter
[18:02:11] <CIA-41> EMC: 03cradek 07master * r46b25f3105ca 10/lib/python/gladevcp/widget-gladevcp-hal_meter.png: gladevcp: Added icon for HAL Meter
[18:02:12] <CIA-41> EMC: 03cradek 07master * r0178a0b0ecaf 10/lib/python/gladevcp/hal_widgets.py: gladevcp: Remove trivial constructors
[18:02:14] <CIA-41> EMC: 03cradek 07master * r804a452b0007 10/src/hal/user_comps/gladevcp.py: gladevcp: Show and reparent window after userfuncs setup
[18:02:16] <CIA-41> EMC: 03cradek 07master * r3957e7f25415 10/src/hal/user_comps/gladevcp.py: gladevcp: Fix halcomp.ready() and cleanup halcmd invokation
[18:02:17] <CIA-41> EMC: 03cradek 07master * r2c9b32525f5f 10/lib/python/ (hal.py hal_glib.py): pyhal: Add GLib wrappers for Pin
[18:02:18] <CIA-41> EMC: 03cradek 07master * r523384fc4e99 10/lib/python/gladevcp/ (hal_bar.py hal_widgets.py makepins.py): gladevcp: Use GPin value-changed signal
[18:02:19] <CIA-41> EMC: 03cradek 07master * r0a74805cc8bd 10/ (3 files in 2 dirs): gladevcp: Wrap glade in gtk.Builder like object
[18:02:32] <CIA-41> EMC: 03cradek 07master * r019aa2b3895c 10/lib/python/gladevcp/ (hal_bar.py hal_widgets.py led.py): gladevcp: Add hal-pin-changed signal to 'output' widgets
[18:02:33] <CIA-41> EMC: 03cradek 07master * r65d7684fb6fc 10/lib/python/gladevcp/ (hal_graph.py hal_python.xml hal_pythonplugin.py): gladevcp: Added HAL Graph widget
[18:02:34] <CIA-41> EMC: 03cradek 07master * rc64c3659496e 10/ (5 files in 2 dirs): Merge branch 'gladevcp-graph' into gladevcp-widgets
[18:02:36] <CIA-41> EMC: 03cradek 07master * r7af2a31c464c 10/lib/python/ (11 files in 2 dirs): Merge branches 'gladevcp-gladebuilder2', 'gladevcp-glib3' and 'gladevcp-widgets' into gladevcp
[18:02:37] <CIA-41> EMC: 03cradek 07master * ra9d5761ad5d6 10/lib/python/gladevcp/hal_graph.py: gladevcp: Added autoscaling to Graph widget
[18:02:37] <CIA-41> EMC: 03cradek 07master * r48dae8cd3b9b 10/lib/python/gladevcp/hal_graph.py: gladevcp: Added snapshotting to graph
[18:02:38] <CIA-41> EMC: 03cradek 07master * r2afcef5e3c90 10/src/hal/user_comps/gladevcp.py: gladevcp: Drop GladePanel argument to user handlers
[18:02:39] <CIA-41> EMC: 03cradek 07master * r309bce42bee2 10/lib/python/gladevcp/persistence.py: gladevcp: Added persistence support
[18:02:40] <CIA-41> EMC: 03cradek 07master * r3943baffa309 10/configs/gladevcp/ (34 files in 8 dirs): gladevcp: Added lot of gladevcp examples
[18:02:44] <CIA-41> EMC: 03cradek 07master * rd16e9d09ff38 10/configs/gladevcp/ (6 files): gladevcp: Added touchy configuration example
[18:02:44] <CIA-41> EMC: 03cradek 07master * rdfe91f2197b1 10/lib/python/gladevcp/hal_meter.py: gladevcp: Provide hal-pin-changed signal for HAL Meter
[18:02:49] <CIA-41> EMC: 03cradek 07master * r8fa3b9b8bd70 10/configs/gladevcp/simpockets.tbl: gladevcp: Added tooltable to touchy sim config
[18:02:57] <CIA-41> EMC: 03cradek 07master * r879a84f2b38c 10/ (56 files in 12 dirs): Merge branch 'gladevcp' of git://psha.org.ru/psha/emc2
[18:03:20] <psha> thanks :)
[18:03:36] <psha> i'll send announce to mailing list shortly
[18:30:24] <psha> skunkworks: you'd mentioned that it would be nice to have graph of following error
[18:30:41] <psha> how fast is it changing?
[18:33:28] <mhaberler_> mhaberler_ is now known as mhaberler
[18:51:00] <skunkworks> psha: every servo cycle
[18:51:37] <skunkworks> I said it would be cool - I don't know if it is needed ;)
[18:53:00] <psha> :)
[19:52:05] <jepler> SWPadnos: re the compile times -- that is interesting
[19:52:47] <SWPadnos> yeah. should have written down the commit I went from, but I didn't
[19:52:53] <jepler> SWPadnos: can you find out what your old ref was? maybe just pastebin the first few lines of 'git log -g --oneline' if you're not sure how to parse it
[19:53:22] <SWPadnos> oh, cool, lemme get that
[19:54:27] <SWPadnos> 514a9bc
[19:55:18] <jepler> let me see if I measure the same here..
[19:57:08] <jepler> by its nature the gladevcp plots are not going to be capturing every servo cycle .. I think replacing halscope is probably not a goal here
[19:57:29] <jepler> my methodology: ./configure --enable-simulator; make clean; make; make clean; time make -j5 (I have 4 cores)
[19:57:52] <SWPadnos> ok, I'll do the same with -j12ish
[19:58:15] <SWPadnos> can you "checkout" a revision like git checkout 514a9bc ?
[19:58:52] <jepler> yes, git checkout 514a9bc
[19:59:22] <psha> jepler: heh, it's not aiming in replacing halscope -- i even don't understand 90% of controls inside scope! :)
[19:59:59] <psha> but with proper scope_rt with ringbuffer it may capture everything
[20:00:45] <jepler> 225.72user 56.89system 0:51.65elapsed 547%CPU (0avgtext+0avgdata 263856maxresident)k
[20:00:58] <jepler> vs 246.93 user in an earlier run
[20:01:15] <SWPLinux> is that report format due to your using debian?
[20:01:17] <jepler> but I think those timings are just a bit suspect; I still have only a 4-core machine
[20:01:29] <jepler> SWPadnos: that's /usr/bin/time which is different from the shell built-in time
[20:01:52] <SWPLinux> yeah, you now appear to have about 3/4 more of a CPU than before
[20:04:35] <SWPLinux> funny how it seems interminably slow doing the initial make, since I didn't use -j on that one
[20:05:34] <SWPLinux> 77.45user 12.52system 0:14.25elapsed 631PU (0avgtext+0avgdata 263952maxresident)k
[20:06:12] <SWPLinux> heh. lost the
[20:06:17] <SWPLinux> heh. lost the
[20:06:19] <SWPLinux> huh
[20:06:28] <SWPLinux> I can't type %%
[20:06:48] <SWPLinux> 78.92user 13.17system 0:14.76elapsed 623%CPU (0avgtext+0avgdata 264096maxresident)k
[20:06:58] <SWPLinux> (need 2 %% signs)
[20:07:06] <SWPLinux> unless there's nothing after it
[20:07:08] <SWPLinux> sigh
[20:07:08] <skunkworks> what does -j5 do?
[20:07:27] <SWPLinux> anyway, about the same now, using /usr/bin/time
[20:07:32] <skunkworks> is it alway one more than the number of cores?
[20:07:36] <SWPLinux> skunkworks: runs multiple processes simultaneously
[20:07:40] <SWPLinux> no
[20:07:43] <SWPLinux> I used 2x
[20:08:33] <skunkworks> ah - thanks
[20:09:18] <jepler> SWPadnos: your system is 6 cores?
[20:09:27] <jepler> .. and you got 623% CPU?
[20:09:33] <jepler> something's verra weird
[20:09:34] <SWPLinux> I probably didn't need to do that in this case. you'd use more than the number of CPUs/cores to allow some processes to wait for I/O
[20:09:36] <SWPLinux> jepler: yes
[20:10:22] <jepler> so in your case user went from about 80s to about 120s
[20:10:45] <jepler> and in my case, 225 -> 250
[20:10:49] <jepler> +50% and +11%
[20:10:54] <SWPLinux> no, they're about the same
[20:11:01] <SWPLinux> 77.45 vs 78.92
[20:11:01] <jepler> oh, hm
[20:11:27] <jepler> 09:42:29 <SWPadnos> user: 1:59.33, sys 0:14.83
[20:11:30] <jepler> that's what you'd said earlier
[20:11:35] <jepler> this time around you got a different measurement?
[20:11:38] <SWPLinux> yes
[20:11:41] <SWPLinux> to both
[20:12:30] <SWPLinux> real0m14.642s
[20:12:32] <SWPLinux> user1m18.970s
[20:12:33] <SWPLinux> sys0m12.660s
[20:12:34] <SWPLinux> using builtin time
[20:12:37] <SWPLinux> so about the same
[20:13:25] <jepler> which command did you use to get back to the current version?
[20:13:34] <SWPLinux> git checkout HEAD
[20:13:45] <SWPLinux> maybe that wasn't right
[20:14:17] <jepler> that's not right
[20:14:21] <jepler> you probably want 'git checkout master'
[20:14:24] <SWPLinux> oh, phew :)
[20:14:58] <skunkworks> wow - fan really spins up when compiling ;)
[20:15:42] <SWPLinux> real0m21.132s
[20:15:44] <andypugh> Does anyone have a complex kinematics setup with a joypad avaialable?
[20:15:44] <SWPLinux> user1m36.690s
[20:15:46] <SWPLinux> sys0m17.240s
[20:16:38] <SWPLinux> 97.93user 17.35system 0:21.39elapsed 538PU (0avgtext+0avgdata 267792maxresident)k
[20:16:46] <SWPLinux> gah
[20:16:51] <SWPLinux> 97.93user 17.35system 0:21.39elapsed 538%CPU (0avgtext+0avgdata 267792maxresident)k
[20:17:12] <SWPLinux> hmm. strange that the CPU usage wasn't even 600%
[20:17:27] <SWPLinux> (relative to the other times being > 100% * core count)
[20:17:34] <andypugh> That chap with the volleyball thing is reporting that he can jog freely in joint mode, but in world mode he can only jog one direction at a time (but he can jog two directions at a time with the keyboard)
[20:17:38] <jepler> SWPLinux: yeah, I'm looking more at user or user+sys than elapsed
[20:18:37] <SWPLinux> ok, so that's a 25% increase or thereabouts
[20:18:38] <skunkworks_> real0m31.284s user3m8.350s sys0m21.090s
[20:19:01] <skunkworks_> (that was the latest trunk)
[20:19:14] <SWPLinux> skunkworks_: how many cores, and what -j number?
[20:19:35] <skunkworks> j5 6 cores (i7)
[20:20:17] <skunkworks> wait - 4 cores (shows 8 cpus)
[20:20:34] <SWPLinux> out of curiosity, try -j10
[20:21:18] <SWPLinux> since you can have 8 threads executing (more or less) simultaneously, that would let there be a couple waiting for something while the other 8 are gobbling up all the CPU time :)
[20:22:35] <skunkworks_> real0m29.517s user4m25.790s sys0m30.570s
[20:23:22] <SWPLinux> huh. it is strange that sys went up by 5 seconds in my case
[20:25:41] <skunkworks_> http://www.electronicsam.com/images/KandT/testing/Screenshot-1.png
[20:25:43] <skunkworks_> :)
[20:26:02] <jepler> I got 225 user seconds at the original commit, 250 seconds at the b8e06ac59 (near origin/master but not quite there), and at 1187c3ca (nearly halfway in-between, the first commit chosen by 'git bisect'), 236
[20:26:29] <jepler> suggesting that there'
[20:26:35] <jepler> s not a smoking gun, just creeping up
[20:26:42] <jepler> .. even though 10% in just a few months is surprising
[20:27:26] <jepler> 283 files changed, 32217 insertions(+), 27564 deletions(-)
[20:28:52] <jepler> the software got 4000 lines bigger over that time?!
[20:29:37] <psha> jepler: todays gladevcp merge added 3.5k lines of examples :)
[20:29:48] <psha> mostly
[20:30:03] <psha> but that won't affect build time
[20:39:43] <mhaberler> the sample glade.ui XML files create a lot of noise
[20:43:04] <skunkworks_> huh - in top - plugin-containe seems to be taking the highest cpu while idleing
[20:44:35] <jepler> skunkworks_: that's the flash ad that you haven't bothered to block
[20:44:52] <SWPLinux> funny. compiz is the highest for me, or firefox
[20:45:17] <SWPLinux> which reminds me. glxgears is very fast on the GTX480, ~103000 FPS
[20:45:47] <psha> psi, chrome and collectd are trying to stay at top here :)
[20:45:51] <SWPLinux> oops, not FPS, in 5 seconds, so ~20600/sec
[20:45:59] <psha> all with 1% cpu in peaks :)
[20:46:00] <skunkworks_> jepler: I guess - I closed firefox and it went away ;)
[20:46:06] <Jymmm> ABP Rules!!!
[20:46:21] <SWPLinux> A Blind Parrot?
[20:47:00] <Jymmm> From FF, just click install.... http://adblockplus.org/en/installation
[20:47:17] <SWPLinux> oh. AdBlock Plus
[20:50:35] <psha> bb
[21:25:11] <ries_> ries_ is now known as ries
[22:02:05] <ries_> ries_ is now known as ries